(296a) Model-Based Diagnosis and Management of Chronic Obstructive Pulmonary Disease (COPD)*

Chronic Obstructive Pulmonary Disease (COPD) is an inflammatory lung disease characterized by a persistent airflow obstruction severe enough to interfere with normal breathing function. The airflow limitation results from bronchial or alveolar abnormalities or a combination of them, usually caused by significant exposure to irritating gases such as wood or cigarette smoke [1,2]. The disease has an estimated global prevalence of over 10% (in the population aged >= 40 years) and is the 3^rd leading cause of death worldwide [2,3]. Consequently, the medical expenditures attributed to COPD are huge—the direct cost in the US alone is projected to approach $40 billion per year in the next few years, with similar expenses in the EU [2,3]. Despite being a significant public health challenge, several impediments exist to recognizing, assessing, and managing COPD, primarily due to the heterogeneity in disease and the patient-to-patient variations in response to therapy [4,5,6].

A key challenge in disease management is the occurrence of exacerbations—worsening of the patient’s condition beyond normal day-to-day variations—which may require a change in medication and/or hospitalization, depending on the severity of the episode. Owing to the hospitalization and ambulatory oxygen requirements, the most significant proportion of COPD-related medical expenses is also attributed to exacerbations [2]. It has been reported that initial symptoms of exacerbations occur well before the actual hospitalization, indicating the possibility of developing warning systems that could predict the occurrence and severity of these exacerbations [7]. In this context, Remote Patient Monitoring (RPM)—using digital technologies to collect medical/health data from individuals in one location and transmit (securely) to healthcare providers in a different location—is considered to hold the potential to improve COPD patient status assessment and treatment recommendations, predict exacerbations, and lower associated healthcare costs. There have been a few recent studies on predicting exacerbations using machine learning models that take as input one or more RPM variables—heart rate, activity level, self-declared state of well-being, spirometry test, and blood work results [7,8,9]. Noteworthy among these is the app-based product COPDPredict^TM, recently released by NEPeSMO, which has been relatively successful [7]. However, the product relies heavily on biomarkers measured from the blood, indicating that the technology is not entirely continuous and requires a human-in-the-loop. In addition, after an initial screening by a decision tree, the diagnosis seems to be performed by a team of physicians rather than the software by itself. Most importantly, this, and none of the other works mentioned above, interprets the RPM data taking into account the physics involved, which can be crucial in predicting the severity of the exacerbation.

We hypothesize that using the RPM data in conjunction with a personalized, physics-driven model of the human cardio-respiratory system could extract more information from the data, enabling RPM to be a continuous process requiring no human intervention, thereby improving the effectiveness of RPM technologies. Specifically, we propose modeling the occurrence of COPD from a control-systems engineering perspective, whereby the cardio-respiratory system—whose physiological functions will be described by appropriate mathematical equations—is represented as a control system (sensor, controller, actuator, process) and the disease state is modeled as a malfunction (or failure) of one or more components of the system. In the past, we have presented results of the basic system model and identified parameters associated with the emergence of COPD [10,11]. In the present work, we will discuss progress on model development and analysis in three parts: (i) the inclusion of the effect of activity level on the human cardio-respiratory control system as a feedforward loop—in addition to the existing feedback loop relying on measurements of arterial gas concentrations (ii) the analysis of the resultant model to understand the dynamics of RPM variables as a function of activity levels, resulting in the crucial conclusion that the RPM variables could encode information on disease states, with worsening disease states driving the RPM variables further from normal, and (iii) with the help of unsupervised learning tools and RPM data from COPD patients, the testing of the hypothesis on whether the disease state can be determined from RPM variables. Based on our analysis of the model—which we believe is the first instance of connecting physiological functions with RPM data in COPD patients—we conclude that RPM, combined with a personalized mathematical model, holds the potential to predict exacerbations, assess its severity, and facilitate timely access to healthcare facilities, thereby enhancing the quality of life for COPD patients.

*The present presentation is dedicated to the memory of the late Prof. Babatunde A. Ogunnaike, a dear colleague, friend and the initiator of this research.

References

[1] Mannino, D. M., & Buist, A. S. (2007). Global burden of COPD: risk factors, prevalence, and future trends. The Lancet, 370(9589), 765-773.

[2] GOLD. (2023). Global Strategy for Prevention, Diagnosis, and Management of COPD: 2023 Report. https://goldcopd.org/2023-gold-report-2/.

[3] World Health Organization. (2022). Chronic obstructive pulmonary disease (COPD). [2023-05-03]. https://www.who.int/news-room/fact-sheets/detail/chronic-obstructive-pulmonary-disease-(copd).

[4] Halpin, D. M., Celli, B. R., Criner, G. J., Frith, P., Varela, M. V. L., Salvi, S., ... & Agusti, A. (2019). It is time for the world to take COPD seriously: a statement from the GOLD board of directors. European Respiratory Journal, 54(1).

[5] Vogelmeier, C. F., Román-Rodríguez, M., Singh, D., Han, M. K., Rodríguez-Roisin, R., & Ferguson, G. T. (2020). Goals of COPD treatment: focus on symptoms and exacerbations. Respiratory medicine, 166, 105938.

[6] Franssen, F. M., Alter, P., Bar, N., Benedikter, B. J., Iurato, S., Maier, D., ... & Schmeck, B. (2019). Personalized medicine for patients with COPD: where are we?. International journal of chronic obstructive pulmonary disease, 1465-1484.

[7] Patel, N., Kinmond, K., Jones, P., Birks, P., & Spiteri, M. A. (2021). Validation of COPDPredictâ„¢: unique combination of remote monitoring and exacerbation prediction to support preventative management of COPD exacerbations. International journal of chronic obstructive pulmonary disease, 1887-1899.

[8] Tiwari, A., Liaqat, S., Liaqat, D., Gabel, M., de Lara, E., & Falk, T. H. (2021, November). Remote copd severity and exacerbation detection using heart rate and activity data measured from a wearable device. In 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) (pp. 7450-7454). IEEE.

[9] Wu, C. T., Li, G. H., Huang, C. T., Cheng, Y. C., Chen, C. H., Chien, J. Y., ... & Lai, F. (2021). Acute exacerbation of a chronic obstructive pulmonary disease prediction system using wearable device data, machine learning, and deep learning: development and cohort study. JMIR mHealth and uHealth, 9(5), e22591.

[10] Ghadipasha, N., Chalant, A., Yu, B., Ogunnaike, B.A. (2019). Model-Based Diagnosis, Management, and Treatment of Chronic Obstructive Pulmonary Disease (COPD). 2019 AIChE Annual Meeting.

[11] Kurian, V., Ghadipasha, N., Gee, M., Chalant, A., Hamill, T., Okossi, A., ... & Beris, A. N. (2022). A Systems Engineering Approach to Modeling and Analysis of Chronic Obstructive Pulmonary Disease (COPD). arXiv preprint arXiv:2212.13207.